1. Database System Architecture
CSCI 6442
©Copyright 2019, David C. Roberts, all rights reserved

2. Agenda Relational and performance Database performance goals
DBMS use of disk
DBMS Architecture

3. Origins of the Relational Approach
First appeared in Codd’s 1970 Communications of the ACM Article
Emphasized data independence
Emphasized more rigorous foundation for data management
Performance of early relational systems so poor that it called into question the practicality of the relational approach

4. Moore’s Law to the Rescue!
1970
1984
1997
2007
2010
Cost
$4,600,000
$4,000
$1,000
$550
$600
Speed (MHz)
12.5
8.3
166
1600
3000
Cost per MHz
$368,000
$482 $6
$.34
$.10

6. What Does This Mean? Expensive Computer Cheap People Cheap Computer
1970
storage and processing resources were scarce and expensive
computer price pays for 400 people for a year
2010
processing and storage are so cheap that they are nearly free
computer price pays for one person for a day
Expensive Computer
Cheap People
Cheap Computer
Expensive People

7. 1970 Data Models Performance the key issue
Data model tailored to a business process
Business process details drive the data model
Code is written for the single business process
Applications can be built to be rather efficient based on such a data model
Correspondence between data model and a single business process is a given

8. The March of Technology
Today’s technology advances make what was brute force and clumsy and expensive yesterday the elegant easy solution for today
We may need to rethink basic approaches in the light of today’s technical economics

9. Relational—Mirroring Previous Systems
Pre-relational: big deal to change the database structure
Database structure was embedded in applications, so applications had to change
Relational: huge improvement for the DBA to be able to change physical structure and not impact applications

10. Relational Performance
Compared with a hand-coded application with custom data structures, a relational database has perhaps 10x poorer performance
We are gaining more efficient use of people and paying for it with cpu cycles

11. Conventional Data Models
Conceptual—model includes entity types, relationships
Logical—model includes entity types, relationships and attributes
Physical—logical model mapped onto physical structures provided by DBMS

12. DBMS Physical Model
DBMS typically stores attributes of a single row near each other
Rows in a table may be in a single file or otherwise co-located
Changing entity types and their attributes in physical database is reserved for the DBA and can’t be done by applications
Why? For performance (a la 1970?), and because earlier DBMSes did it that way

13. What a DBMS Does At its heart, a RDBMS offers three things:
Tables
Attributes of tables
Constraints
Application code is written to do CRUD operations on tables and enforce constraints
Transaction processing and access control are built in to the DBMS
Other important features are built on these basic ones

14. Components of a DBMS

15. DBMS Architecture Data is stored on disk
Disk is necessary for database to be reliably available
Disk is millions of times slower than anything that happens in RAM
Number of disk accesses is a good measure of DBMS cost for an operation

16. Disk
Disk is composed of fixed-length records, rotating around
To access information, we need to move the head and wait for the disk to rotate
We wait the same time whether we use one byte or all the record
We call this fixed length record a page

17. Efficient Use of Disk
For efficient use of disk, we want to use all the information contained in a single page
We will look at how we organize disk in order to reduce the number of disk accesses for a search

18. Disk vs. RAM RAM is accessible in any order
Any sort of structures can be used
Data structure courses usually cover data structures for RAM
We’ll talk about how to make efficient use of disk

19. Disk as Pages
Disk is composed of fixed-length records, rotating around
To access information, we need to move the head and wait for the disk to rotate
We wait the same time whether we use one byte or all the record
We call this fixed length record a page

20. Physical Implementation
The DBMS gets a file from the OS that it then writes
One or more of these files are managed as the database
The DBMS allocates space within physical records to use a rows and index blocks
A database row is implemented as a logical record in the file system
Thus, the DBMS actually physically implements the row structure of the database
Which has the same entity types as the conceptual data model
Which has the same attributes as the logical data model

21. The Database
Extent 1
The database may be spread across multiple physical disk drives
Extent 2
Row
Extent 3
<<tid>,<rid>,<cid><cli><cv>, … ,
<cid>,<cli>,<cv>, … >>

22. DBMS and Applications Database Management System Application Buffer
Program
Buffer
Database
Management
System
Application
Program
Buffer
Application
Program
Buffer
Application
Program
Buffer

23. DBMS Software Architecture
Application
Program
Buffer
System
Global
Area
Database
System
Application
Program
Buffer
Application
Program
Buffer
Application
Program
Buffer

24. SQL Processing Lexical Analyzer Syntax Analyzer Executor Results SQL
Tokens
Syntax
Analyzer
Quads
Executor
Results

25. Executor Software Architecture
SQL Executor
Table Management
Index Management
Row Management
Node Management
Page Management
Data Store

26. Question: is there a third kind of page?
Pages
Disk is divided into physical records called “pages”
A page can be an index page (ie b-tree) or a data page
Index page contains one node of a b-tree
Data page contains rows of tables
Question: is there a third kind of page?

27. Page Allocation Pages are initially considered all unallocated
In response to requests, they are allocated and marked allocated
When freed, they are chained onto a list of free pages

28. Database Extents
Database needs to be able to extend over disk boundaries
Size may require it
Growth may require it
Typically it’s managed as “extents”, each of which is a file to the OS file system
Multiple files are mapped into a single sequence of page IDs

29. Extents SQL Executor Table Management Index Management Row Management
Node Management
Page Management
Extent Management
Data Store

30. The Database Extent 1 Extent 2 Row Extent 3
<<tid>,<rid>,<cid><cli><cv>, … ,
<cid>,<cli>,<cv>, … >>

31. Startup At startup, DBMS creates an empty system catalog
Catalog has images of some tables; once images are established, then SQL can be used to create other tables

32. System Catalog
The DBMS uses the system catalog to track objects of interest
When the DBMS starts with a new database, it lays down part of the system catalog from an image
The rest of the system catalog is created by SQL statements
Many SQL statements reference or change the system catalog

33. System Catalog
You will have the opportunity to learn more about the system catalog in your assignment.